The sodium, potassium, and water contents of red cells of healthy human adults.

L J Beilin, … , A D Munro-Faure, J Anderson

J Clin Invest. 1966;45(11):1817-1825. https://doi.org/10.1172/JCI105485.

Research Article

Find the latest version: https://jci.me/105485/pdf Journal of Clinical Investigation Vol. 45, No. 11, 1966

The Sodium, Potassium, and Water Contents of Red Blood Cells of Healthy Human Adults *

L. J. BEILIN,t G. J. KNIGHT, A. D. MUNRO-FAURE,4 AND J. ANDERSON (From the Department of Medicine, King's College Hospital Medical School, and the Medical Research Department, the Wellcome Foundation Ltd., London, England)

Intracellular sodium and potassium concentra- coat-RBC interface by an electric saw.4 The cut cen- tions in man can only be measured with ease and trifuge tube containing red blood cells was transferred to a glass beaker and weighed. The weight of red blood accuracy in red blood cells (RBC). Although cells sampled was obtained by reweighing the beaker and there are many published reports of RBC sodium cut centrifuge tube after washing clean and oven drying. and potassium concentrations, the number of Plasma and RBC water content was estimated by healthy subjects studied has been small, indirect washing the contents of the beakers into weighed plastic methods of measurement have often been used, Petri dishes, which were dried at 450 C to constant and biological sources of variation have not been weight. Plasma and RBC samples for sodium and po- tassium estimation were diluted to appropriate concentra- defined. This paper describes the results of a tions and stored in polythene containers at 40 C for 1 to study of 142 healthy adults and delineates some of 4 days before analysis. the factors influencing RBC sodium and potas- All water used for washing equipment and diluting sium concentrations. had an electrical resistance of 106 ohms. Sodium and potassium concentrations were determined with a flame photometer 5 with standards made from Methods P.V.S. sodium chlorides (an especially pure grade for Blood was drawn with minimal venous compression volumetric standardization) and analytical grade potas- into plastic syringes containing ammonium heparin' in sium chloride.6 Two plasma reference standards (Versa- isotonic (0.005 ml of solution containing 0.25 mg tol7 and Wellcome8) were routinely used. Potassium heparin per ml of ). Five-g samples of blood and , at concentrations found relative to sodium in red were immediately centrifuged in cellulose nitrate tubes blood cells, did not influence the sodium flame. Plasma (i.d. 11 mm) 2 for 1 hour in a governed cen- potassium was estimated against a standard containing trifuge.3 The relative centrifugal force (RCF) at the sodium. base of the tubes was 1,460 g; mean RCF applied to the was estimated (1) by adding 1 ml of the cells thus depended on hematocrit and varied from ap- hemolysate prepared for sodium determination to proximately 1,330 to 1,370 g. Two or three tubes were 10 ml of a modified Drabkin's reagent, prepared by dis- used for estimation of plasma and RBC sodium and po- solving one Aculute9 pellet in 250 ml water. After 15 tassium concentrations and one or two tubes for water minutes the optical density at 540 mlu of the cyanmethe- and hemoglobin content. moglobin solution was compared in a spectrophotometer 10 After spinning, a weighed plasma sample was removed with that of a reference standard 11 that conformed to and the centrifuge tubes containing a layer of plasma and the International Standard. the packed cell column were frozen in an ethanol-CO2 The method of estimating plasma sodium trapping, mixture and sectioned 0.5 to 1.0 mm below the buffy which is fully described elsewhere (2), is based on the use of sucrose-J"C and 'Na as plasma markers. This * Submitted for publication April 7, 1966; accepted method was developed when it was found that iodinated August 12, 1966. human (IHSA) underestimates the The contents of this paper formed part of a thesis by amount of trapped extracellular sodium in the packed Dr. L. J. Beilin for the degree of M.D. of the Uni- versity of London. 4Apparatus specially constructed by the Department of t Present address: Dept. of Medicine, Hammersmith Physics, King's College Hospital. Hospital, Du Cane Road, London, W. 12, England. 5Evans Electroselenium, Harlow, England. t: Address requests for reprints to Dr. A. D. Munro- 6 Hopkins & Williams, Chadwell Heath, England. Faure, King's College Hospital Medical School, Den- 7William Warner, Eastleigh, England. mark Hill, London, S. E. 5, England. 8 Burroughs Wellcome, London, England. 1 Evans Medical, Liverpool, England. 9 Ortho Pharmaceutical, Saunderton, England. 2International Equipment, Needham Heights, Mass. 'OOptica Cf4, Baird & Tatlock, London, England. 3 Measuring & Scientific Equipment, London, England. 11 C. Davis Keeler, London, England. 1817 1818 BEILIN, KNIGHT, MUNRO-FAURE, AND ANDERSON cell layer. Plasma trapping in the packed RBC column In the second and third groups of subj ects, water con- was obtained for each sample from the regression equa- tent, and in the third group, cell hemoglobin concentration, tion y = 2.8214 + 0.2398 x, where y is the percentage were also measured. trapping of plasma and x the weight in grams of red blood cells sampled. Results Red cell sodium, potassium, and water contents were corrected for trapped plasma sodium, potassium, and wa- Subject-to-subject and day-to-day variation ter by application of the following formula: Corrected RBC content = (observed RBC content - T X plasma Red cell sodium and water contents were mea- content) /1 - T, where T is the fraction of plasma trapped sured in six men and four women and potassium in the packed cell layer. Results were expressed in content in three men and three women, on three terms of and wet water, and plasma RBC weight, solids. to seven occasions over a period of 9 months. Calculatjons were performed by digital computer, The results are listed in Table I. It can be seen which was also used for many of the statistical analyses. Standard statistical techniques were applied. that RBC sodium concentration was relatively con- The standard errors of the mean of two blood samples stant in individual men and women throughout (from 12 subjects) were as follows: for plasma sodium, this period. Subject-to-subject variation was sig- potassium, and water, respectively, 0.25 mEq, 0.10 mEq, nificantly greater than the day-to-day variation for and 0.21 g per kg plasma, and for RBC sodium, potas- an individual (p < 0.001). sium, water, and hemoglobin, respectively, 0.08 mEq, 0.42 mEq, 0.16 g, and 2.6 g per kg RBC. Subject-to-subject variation of RBC potassium Three groups of subjects were studied. All were concentration was also significantly greater than healthy and at work and none was taking medication. the day-to-day variation (p < 0.001), and day-to- Sodium and potassium concentrations only were mea- day variation was significantly greater in women sured in the first group, who were studied sequentially than in men (p < 0.05). When the values in over a period of 2 months. A significant correlation women were related to days of the menstrual cycle, was found between potassium concentration (in both plasma and red blood cells) and the date of sampling. they suggested that potassium concentration in red Because this correlation was probably due to technical blood cells falls during the cycle. Thus the aver- factors, only the sodium results from this group were age change in potassium concentration from the used. highest levels (mEq per kg cells) was as follows: The samples in the second and third groups of subjects were collected, processed, and analyzed as groups. Be- day 3, 0; day 10, -2.3; day 19, -2.97; and day cause analysis of the results from the first group sug- 26, - 3.94. These results were not obtained dur- gested the existence of age and sex effects, the second ing single cycles. group was balanced for these factors, and in the third For RBC water content, subject-to-subject vari- group they were eliminated. The sex and age composi- tion of the groups is shown in Table II. ation was no greater than day-to-day variation.

TABLE I The sodium, potassium, and water contents of red blood cells from ten subjects estimated on three to seven occasions to illustrate variation among subjects and among occasions*

Subject: 1 2 3 4 5 6 7 8 9 10 Sodium, 6.09 10.52 6.89 5.53 7.55 5.15 7.19 5.94 8.25 4.57 5.06 mEq/kg 6.44 11.13 6.94 6.19 6.71 5.18 7.22 6.36 8.30 5.20 5.07 cells 6.02 10.61 7.23 5.57 6.51 5.20 7.33 6.24 8.04 4.76 4.49 6.71 4.68 Potassium, 92.52 85.69 87.01 93.01 93.83 98.30 mEq/kg 91.52 85.05 87.64 92.16 89.52 96.23 91.96 85.33 87.12 92.30 91.81 91.97 84.85 98.73 Water, 642.5 637.8 634.0 645.2 640.9 631.8 640.6 643.3 655.0 632.6 649.5 g/kg 629.5 637.7 625.2 630.0 641.9 651.8 651.4 651.4 644.2 677.3 651.4 625.5 638.0 624.1 637.0 642.1 651.4 618.2 649.7 652.0 642.6 652.0 656.5 654.5

* Each value given is the mean of two or three estimations made on a single occasion. The average interval between occasions was 31 days (range, 8 to 203 days). Subjects 1 to 6 were men; subjects 7 to 10 were women. RED CELL SODIUM AND POTASSIUM 1819

TABLE II Mean sodium and potassium concentrations and water content of red blood cells*

Age (years): Age Sex Group 20-25 26-35 36-45 46-55 56-65 effect effect Pooled data Na, mEq/kg cells Men 1 7.510 ( 8) 7.897 (7) 7.287 (3) 7.658 (11) 8.013 (4) NS 2 7.710 (12) 7.122 (12) NS 7.446 (87) 3 7.174 (30) p <0.01 Women 1 6.669 (10) 6.354 (2) 7.114 (12) 7.397 (7) P <0.10 <45 years >45 years 2 6.332 (12) 7.361 (12) p <0.05 6.500 (24) 7.304 (31) P <0.01 K, mEq/kg cells Men 2 88.46 (12) 89.71 (12) NS 88.42 (54) 3 88.23 (30) p <0.001 Women 2 92.20 (12) 92.64 (12) NS 92.41 (24) H2O, g/kg cells Men 2 638.4 (12) 634.7 (12) NS 636.1 (54) 3 635.7 (30) p <0.05 Women 2 638.5 (12) 649.3 (12) p <0.05

* Age and sex effects are summarized. The number of subjects contributing to each mean is given in parentheses. Where appropriate, results have been pooled; these are listed on the right. p values for age effects were derived from correlation coefficients and for sex differences from I tests. The standard deviations for each mean in the pooled results, based on the largest number of degrees of freedom available, were as follows: for sodium, 1.246 mEq per kg cells; for potassium, 2.863 mEq per kg; and for water, 11.22 g per kg cells. Day-to-night variation water content of red blood cells of men, but the Blood samples were collected from six men be- cells of older women contained more water than tween 12 midday and 6 p.m.; a further sample those of younger women (p < 0.05) or of men from each subject was collected 12 hours after the (p < 0.05). first. Analysis of variance showed no significant The second group of subjects was balanced for differences of sodium, potassium, or water con- age and sex, and the results obtained from it were tents of the cells, attributable to the times of sam- suitable for analysis of the variance attributable pling. Mean values for the six subjects (per kg to these two factors. Such analysis showed that RBC) were these: sodium content by day, 8.151 sodium content in terms of RBC dry weight was mEq, and at night, 8.164 mEq; potassium content lower in young women than in young men (p < by day, 86.46 mEq, and at night, 85.13 mEq; wa- 0.001) or in middle-aged women or men (p < ter content by day, 651.0 g, and at night 651.0 g. 0.001). The time of blood sampling was recorded in all Thus the water content of red blood cells was the studies subsequently described. No similar in young men and women, but the sodium correlation content lower in was observed between any of the parameters and was young women. The sodium the time of blood sampling. concentration in cell water was therefore lower in young women than in men. Analysis of the The of age and sex derived data confirmed this (p < 0.01). In effects middle age, the water content of red blood cells The effects of age and sex on RBC sodium, po- of women was greater than in youth, but sodium tassium, and water contents are summarized in content was still greater, and sodium concentra- Table II. tion in cell water was similar in the two sexes. Red cell sodium concentration. Age did not af- Because the age distribution of women studied fect RBC sodium concentration in men, but in was bimodal, it was not possible to determine women concentration in middle age was higher whether the rise of RBC sodium concentration was than in youth (p < 0.01). Values for middle-aged a gradual process from youth to middle age or women were similar to those for men, but the cells whether it was localized and occurred more rapidly of young women contained less sodium than those at the time of the menopause. Although the men- of men (p < 0.01). strual status of all subjects was recorded, the Red cell ewater content. Age did not affect the numbers of pre- and postmenopausal women be- 1820 BEILIN, KNIGHT, MUNRO-FAURE, AND ANDERSON

TABLE III Significant correlations between the parameters measured

Correlation coefficients Group 2 Group 3 Variates Reference weight Men Women Both sexes Men RBC* hemoglobin Sum of RBC Na + K Whole cells -0.5741f RBC hemoglobin RBC water Whole cells -0.4560$ RBC potassium RBC water Cell solids 0.7792t 0.8073t 0.8112t 0.8908t Sum of RBC Na + K RBC water Cell solids 0.8057t 0.8463t 0.8447t 0.9331t RBC sodium RBC potassium Cell water -0.3112 -0.2232 -0.3461§ -0.6007f Plasma water RBC water Plasma and RBC solids 0.1665 0.4946: 0.3903$ 0.1433

* RBC = red . f P <0.001. tp <0.01. § p < 0.05. tween the ages of 46 and 55 were too small to nificantly lower than the mean value of 904.35 allow worthwhile comparison. 3.96 g per kg plasma found in middle-aged men Red cell potassium concentration. No signifi- (p < 0.05). In women the water content of cant correlation between RBC potassium concen- plasma did not alter significantly with increasing tration and age was found, either for men or for age; the mean value was 904.66 + 4.4 g per kg women. At all ages, potassium concentration was plasma. Two factor analysis of variance showed greater in the cells of women than in those of men. that sodium concentration in plasma water was These differences were significant whether the significantly higher in men than in women (p < values were expressed in terms of cell weight (p < 0.01) and in youth than in middle age (p < 0.01). 0.001), of cell water (p < 0.01), or of cell solids Plasma potassium concentration was not in- (p < 0.001). fluenced by age or sex. The mean value was 4.15 Plasma sodium, potassium, and water contents. 0.32 mEq per kg plasma. Mean sodium concentrations per kg plasma in the Red cell hemoglobin concentration. Hemoglo- subjects of group 2 were as follows: for young bin concentration was measured in the third group men, 141.1 + 1.98 mEq; for middle-aged men, of 30 young men. Mean concentration was 342.6 138.4 + 3.04 mEq; for young women, 138.5 + 17.07 g per kg cells, or 8.36 millimolal (assum- 1.86 mEq; and for middle-aged women, 137.4 ing a molecular weight for hemoglobin of 64,458). 2.11 mEq. Mean water content was 901.45 + Interrelationships. Correlation matrixes re- 3.96 g per kg plasma in young men and was sig- lating each parameter to all the others were con-

TABLE IV Regression equations relating the parameters measured*

% vari- ation of dependent Equa- variable Dependent Reference tion accounted variable weight no. Regression equation for p H20, g/kg Cell solids 1 H20= - 268.1 + 23.33 Hgb + 9.942 Na + 6.104 K 91.22 <0.001 H20, g/kg Cell solids 2 H20= - 250.6 + 23.74 Hgb + 6.302 Na + 6.302 K 90.02 <0.001 K, mEq/kg Cell solids 3 K = 61.96 + 0.002631 H20 -1.427 Na - 3.215 Hgb 89.35 <0.001 Na + K, mEq/kg Cell solids 4 Na + K = 60.42 + 0.002550 H20- 3.142 Hgb 89.37 <0.001 Na, mEq/kg Cell solids 5 Na = 7.793 + 0.000791 H20- 0.2589 K 43.87 <0.001 Na, mEq/kg Cell water 6 Na = 47.78 -0.2630 K 36.08 <0.001 K, mEq/kg Cell water 7 K = 154.2 -1.372 Na 36.08 <0.001

* All independent variables are expressed as milliequivalents or millimoles per kilogram. In equation 2, sodium and potassium have been constrained to exert the same effect. The variance ratio F attributable to this constraint was 3.521 and was not significant (0.10 > p < 0.05). Hgb = hemoglobin. RED CELL SODIUM AND POTASSIUM 1821 structed for each group of subjects, with the sexes young cells at the top of the column contain more separated and combined. Significant interrela- potassium than older cells at the bottom (9). tionships are listed in Table III. Mean values for RBC water content found in 'Quantitative relationships among the hemo- this study are compared with other reported values globin, water, sodium, and potassium contents of in Table V. General agreement is good, but lower red blood cells were further investigated by per- levels were found in this study than in the others; forming multiple regression analyses on the data of these Valberg and his associates (3) report the (in terms of cell dry weight and of cell water) highest levels. Two factors may be partly re- from the 30 young men of group 3, with each sponsible for this difference. First, we used a variable in turn dependent on the other three. lower temperature for drying. This temperature If any of the three independent variables failed to was selected because decomposition and volatiliza- contribute significantly (p < 0.05) to the varia- tion of organic material appeared to occur at tion of the dependent parameter, it was rejected 1000 C. Second, the mean hemoglobin content of and the regression recomputed. Thus when he- the cells may have been greater in the subjects of moglobin content was made the dependent vari- this study than in the subjects of the other re- able, all three independent variables (sodium, po- ported series. Valberg and his colleagues, for in- tassium, and water contents) were rejected, indi- stance, found a mean hemoglobin concentration of cating that variation of hemoglobin content could 289 g per kg cells in the subjects they studied. not be explained in terms of the other parameters The concentration found in the 30 young men of measured. The regression equations, the percent- group 3 was 342 g per kg cells. Since hemoglobin ages of the variation of the dependent variables forms such a large proportion of the cell solids, a accounted for by the independent variables, and negative correlation between hemoglobin and wa- the statistical significance of the findings are listed ter content of the cells can be anticipated and was in Table IV. in fact found (Table III). It is probable that some of the variation of RBC water content in dif- Discussion ferent published reports is thus attributable to Red cell sodium concentrations found in this variation of hemoglobin content in the cells of the study are lower than those given in most recent subjects studied. reports. This is due to the use of a different Potassium is the principal intracellular cation, method for estimating trapped plasma sodium. A and a positive correlation between the potassium description of the method used and a critical ex- and water contents of cells can also be anticipated. amination of results obtained by other methods are A close correlation was found, the r rising in the given by Beilin,-Knight, Munro-Faure, and An- third and most homogeneous of the groups to derson (2). 0.8908, a value similar to that noted for the cor- Red cell potassium concentrations found in this relation between the sodium and water contents study are similar to several previous estimates. of plasma (0.8459). When sodium content was Most of the recently published results are given in added to the potassium content, the r between the Table VII of the paper by Valberg, Holt, Paulson, combined cation and the water content of the cells and Szivek (3) and need not be duplicated here. was 0.9331. That 87%o of the variation of cell wa- could be Part of the variation of the reported results can ter content of this group of 30 subjects variations of the sodium and be attributed to one or more of the following fac- explained by potas- sium contents of the cells to if he- tors: derivation of RBC potassium concentra- (rising 91%o moglobin concentration was also included), is a tion from measurement of whole blood and plasma reflection of the constancy of the plasma osmolari- potassium concentrations (4, 5); failure to al- ties of this group. It can be noted in Table III low for plasma trapped in the red cell column that RBC water content was influenced to a trivial (6); the use of hospitalized patients as normal extent in men by the plasma water content, but control subjects, particularly if some are anemic that in the group of women studied the plasma (7, 8); and the use of a part only of the packed water content exerted a significant effect. cell column for potassium estimation (3), since The regression equations for the water content 1822 BEILIN, KNIGHT, MUNRO-FAURE, AND ANDERSON

TABLE V Estimates in the literature of water content*

Mean Temperature water Authors No. of subjects Method of oven content °C g/kg cells Maizels (7) 6 patients Not stated Not stated 648 (some anemic) Hutt (8) 14 patients Indirect 100 652 (some anemic) Nichols and Nichols (10) 21 Indirect Not stated 658 Keitel, Berman, Jones, and 11 men Direct 105 660 MacLachlan (11) 11 women Kessler, Levy, and Allen (4) 24 Indirect 105 650 Czaczkes, Ullmann, Ullmann, and Bar-Kochba (12) 20 Direct 85 648 Valberg and associates (3) 50 Direct 105 681 Funder and Wieth (13) 61 men Direct 105 664 67 women Diet15667 Present study 54 men 636 24 women (20-25 years) Direct 45 638 12 women (46-55 years) 649 * Results given in terms of cell volume have been divided by 1.096 [Ponder (14)]. When red blood cell water con- tent has been estimated from the difference between the contents of whole blood and plasma; it is termed "indirect"; when estimated in red blood cells themselves, it is termed "direct." of the cells (no. 1 and 2, Table IV) are of some obtained from the virial equation of Dick and interest. The size of the regression coefficient b Lowenstein (16), which gives a value of 3.29 at a for each of the independent variables reflects the hemoglobin concentration of 8.36 millimolal. This osmotic weightings of the variables. equation was based on the results of Adair (17), The values for b in equation 1 suggest that he- who studied the osmotic pressures of pure hemo- moglobin exerts an osmotic effect 3.82 times globin solutions at concentrations up to 7.7 milli- greater than that exerted by potassium. In this molal. small group of subjects the coefficients b for so- It is difficult to relate these results to those of dium and potassium do not differ significantly, and Savitz, Sidel, and Solomon (18), who were un- in equation 2, where they have been constrained to able to explain RBC volume changes produced exert the same effect, hemoglobin has an osmotic in solutions of varying in terms of a effect 3.78 greater than that of the combined ca- changing osmotic coefficient for hemoglobin. tions. If it is assumed that the osmotic coefficient These authors, however, measured cell volume of potassium is 0.95 at the observed mean hemo- from the hematocrit, corrected for trapped plasma globin concentration of 8.36 millimolal, the os- by estimation of IHSA trapping. Beilin and his motic coefficient of hemoglobin at this concentra- associates (2) have shown that IHSA underes- tion becomes approximately 3.6. timates trapped plasma sodium and hence prob- The figure of 3.6 at 8.36 millimolal is slightly ably trapped plasma water. Red cell water con- lower than the values obtained by McConaghey tent may thus have been spuriously elevated in the and Maizels (15) in in vitro studies of red blood calculations of Savitz and his colleagues. cells, depleted of cations by incubation in lactose Although cells of low hemoglobin content must media or suspended in saline solutions of varying contain more water and hence more potassium tonicities. The size of their published graphs and sodium, in biological terms the water content does not allow accurate interpolation, but it is of the cells is a function of the cation content doubtful whether the osmotic coefficients differ (assuming constant external osmolarity) and not significantly. vice versa. The forces controlling the cation and Conversely, the figure of 3.6 is higher than that hence the water content of cells have not been RED CELL SODIUM AND POTASSIUM 1823 clearly defined, although models have been pro- and hemoglobin concentration have been defined, posed and tested (19). it is not possible to evaluate the regressions ob- Maizels (7) first noted that potassium concen- served in greater detail. tration in the red blood cells of patients with hy- The results of this study show that red blood pochromic is greater than in the cells of cells of young women contain more potassium and normal subjects. The same inverse relationship less sodium than the cells of young men. Love between potassium and hemoglobin concentrations and Burch (22) found similar differences in potas- was noted in the healthy young men of the third sium but not in sodium concentration. Conversely, group (r, - 0.4821). The r was greater Keitel and his colleagues (11), Dowben and Hol- (- 0.5741) when sodium and potassium together ley (23), and Funder and Wieth (13) noted simi- were related to the hemoglobin content of the cells. lar differences in sodium but not in potassium The regression equations relating these variables concentration. Czaczkes and co-workers (12) (no. 3 and 4, Table IV) indicated that there was and Valberg and associates (3) were unable to a reciprocal change of approximately 3 mEq cation detect any sex effect. per 1 mmole change of hemoglobin content of the It is possible that the higher potassium content cells. This relationship was only apparent when of red blood cells of women than of men is related quantities were expressed in terms of cell weight to a lower hemoglobin content; the latter was not or weight of cell solids. Cation concentration in measured in women in this study. Nevertheless, cell water is a function of plasma osmolarity and is significant sex differences remain when con- independent of hemoglobin concentration. This centrations are expressed in terms of cell water, relationship may be due to increased binding of and these appear to be independent of hemoglobin cation by hemoglobin, as the concentration of the concentration. latter falls over the observed range. The ratio of intracellular potassium to sodium It has been mentioned that the concentrations concentration is about 20% greater in young of sodium and potassium in cell water were inde- women than in young men (14.18 to 11.88). pendent of hemoglobin concentration; they were, In men RBC cation concentrations are not af- however, negatively correlated with each other. fected by increasing age, but in women both the From the two regression equations (no. 5 and 6 sodium and water contents of the cells are greater of Table IV), the change in potassium concentra- in middle age than in youth. Funder and Wieth tion for a 1 mmole change of sodium concentration (13) also found that sodium content, but not wa- could be determined and was found to be between ter content, of red blood cells increases with age 1.37 and 3.8 mmoles. Over the much greater in women. The net effect of these changes is that range of RBC cation concentrations occurring in the sodium concentration in cell water rises from sheep, Tosteson and Hoffman (19) found that the a mean value of 10.18 millimolal at 23 years to change in potassium concentration for a 1 mmole 11.24 millimolal at 50 years, whereas potassium change of sodium concentration was approximately falls from 144.3 to 141.19 millimolal during the 1 mmole. same period. It is probable that hormonal factors The application of compartmental analysis to are responsible for this change with age in women in vivo studies of sodium flux between plasma and and possible that the same factors cause the sex red blood cells (20) has shown that RBC sodium difference. The observation that red blood cell is distributed in at least two compartments. Con- potassium concentration falls during the menstrual sidered as compartments in series, 25 to 35% of cycle requires confirmation in studies during single RBC sodium is in a slowly exchanging inner cycles. compartment. Similar studies with potassium Although some of the factors influencing RBC (21) indicate that this ion is also distributed in at sodium and potassium concentrations have been least two compartments, but suggest that a much indicated in this study, much of the variation greater proportion, approximately 90%, is present among subjects remains unexplained. Thus in the in an inner slowly exchanging compartment. Un- third and most homogeneous groups of subjects, til the relationships between the rapidly and slowly only one-third of the observed variation among exchanging sodium and potassium compartments subjects was accounted for. In the second group 1824 BEILIN, KNIGHT, MUNRO-FAURE, AND ANDERSON of subjects, ABO blood groups were determined, lege Hospital and the staff of the Wellcome Foundation, and the results suggested that ion concentrations who volunteered to take part; to them and to Mr. David Beaumont, who provided much technical assistance, we the cells were influenced by blood group within owe a debt of thanks. substances. This finding was not confirmed in the We also wish to thank Dr. T. E. Cleghorn, Dr. third group of subjects, which was balanced for M. J. S. Langman, Professor H. Harris, and Dr. Eliza- phenotypes A and 0. No clear relationship was beth Robson for their assistance and the Wellcome Foun- Council for found in this group between intracellular ion con- dation and the Medical Research generous support. centrations and blood groups, , or any particular isoenzyme of acid phosphatase, References phosphoglucomutase, 6-phosphogluconate dehy- 1. Dacie, J. V., and S. M. Lewis. Practical Haematol- drogenase, or adenylate kinase. ogy, 3rd. ed. London, J. and A. Churchill, 1963. 2. Beilin, L. J., G. J. Knight, A. D. Munro-Faure, and Summary J. Anderson. The measurement of sodium con- centration in human red blood cells. J. gen. Phys- 1. Red blood cell sodium and potassium concen- iol. 1966, 50, 61. trations were measured in groups of healthy adults 3. Valberg, L. S., J. M. Holt, E. Paulson, and J. comprising 142 subjects. Szivek. Spectro-chemical analysis of sodium, po- tassium, calcium, magnesium, copper, and zinc in 2. For both sodium and potassium concentra- normal human erythrocytes. J. dlin. Invest. 1965, tions subject-to-subject variation was greater than 44, 379. day-to-day variation in an individual. No sig- 4. Kessler, E., M. R. Levy, and R. L. Allen, Jr. Red nificant day-to-night variation was observed. cell in patients with edema. J. Lab. 3. Red blood cell sodium concentration was dlin. Med. 1961, 57, 32. 5. Singer, M. M., H. R. Hoff, S. Fisch, and A. C. De- young women (mean 6.50 mEq per kg lower in Graff. Red blood cell potassium. Therapeutic cells) than in men (mean 7.45 mEq per kg cells) implications. J. Amer. med. Ass. 1964, 187, 24. and increased with age (to a mean 7.30 mEq per 6. Losse, H., H. Zumkley, and H. Wehmeyer. Unter- kg cells at 50 years). No change with age was ob- suchungen fiber den Elektrolyt- und Wassergehalt served in men. Red blood cell potassium con- von Erythrozyten bei arterieller Hypertonie. Z. Kreisl.-Forsch. 1962, 51, 43. centration was greater at all ages in women (mean 7. Maizels, M. The anion and cation contents of nor- 92.41 mEq per kg cells) than in men (mean 88.42 mal and anaemic bloods. Biochem. J. 1936, 30, mEq per kg cells). Age did not affect the con- 821. centration in either sex. 8. Hutt, M. P. Effect of disease on erythrocyte and 4. Ninety-one per cent of the observed variation plasma potassium concentrations. Amer. J. med. of water content of the red blood cells of young Sci. 1952, 223, 176. in men was accounted for by the observed variation 9. Bernstein, R. E. Alterations metabolic energetics and cation transport during aging of red cells. J. of sodium, potassium, and hemoglobin contents dlin. Invest. 1959, 38, 1572. of the cells. In the women studied, plasma water 10. Nichols, G., Jr., and N. Nichols. equi- content also affected red blood cell water content. libria in erythrocytes during diabetic acidosis. J. The osmotic coefficient of hemoglobin at the con- dlin. Invest. 1953, 32, 113. centrations found was 3.8 times greater than the 11. Keitel, H. G., H. Berman, H. Jones, and E. Mac- coefficients for sodium and potassium. Lachlan. The chemical composition of normal hu- man red blood cells, including variability among 5. Sodium and potassium concentrations in red centrifuged cells. Blood 1955, 10, 370. blood cell water were negatively correlated. The 12. Czaczkes, J. W., T. D. Ullmann, L. Ullmann, and Z. regression equations relating these variables indi- Bar-Kochba. Determination of the red blood cell cated that for a given change of sodium concentra- content of water, sodium, and potassium in nor- tion the change of potassium concentration was 1.4 mal subjects. J. Lab. dlin. Med. 1963, 61, 873. to 3.8 times as great. 13. Funder, J., and J. 0. Wieth. Potassium, sodium, and water in normal human red blood cells. Scand. J. dlin. Lab. Invest. 1966, 18, 167. Acknowledgments 14. Ponder, E. The relation between red blood cell den- This study could not have been undertaken without sity and corpuscular hemoglobin concentration. J. the co-operation of the medical students of King's Col- biol. Chem. 1942, 144, 333. RED CELL SODIUM AND POTASSIUM 1825

15. McConaghey, P. D., and M. Maizels. The osmotic and low potassium sheep red cells. J. gen. Phys- coefficients of haemoglobin in red cells under vary- iol. 1960, 44, 169. ing conditions. J. Physiol. (Lond.) 1961, 155, 28. 20. Beilin, L. J., D. Eyeions, G. Hatcher, G. J. Knight, 16. Dick, D. A. T., and L. M. Lowenstein. Osmotic A. D. Munro-Faure, and J. Anderson. The en- equilibria in human erythrocytes studied by im- try of sodium into human red blood cells in vivo. mersion refractometry. Proc. roy. Soc. B 1958, J. gen. Physiol. 1966, 50, 75. 148, 241. 21. Beilin, L. J., G. J. Knight, A. D. Munro-Faure, and 17. Adair, G. S. The thermodynamic analysis of the J. Anderson. Unpublished results. observed osmotic pressures of salts in solu- 22. Love, W. D., and G. E. Burch. A comparison of tions of finite concentration. Proc. roy. Soc. A potassium', rubidium", and cesium' as tracers of 1929, 126, 16. potassium in the study of cation metabolism of hu- 18. Savitz, D., V. W. Sidel, and A. K. Solomon. Os- man erythrocytes in vitro. J. Lab. clin. Med. 1953, motic properties of human red cells. J. gen. 41, 351. Physiol. 1964, 48, 79. 23. Dowben, R. M., and K. R. Holley. Erythrocyte 19. Tosteson, D. C., and J. F. Hoffman. Regulation of electrolytes in muscle disease. J. Lab. clin. Med. cell volume by active cation transport in high 1959, 54, 867.